Measurements*of strongphaseinD Kπdecayandy...

Miami 2013 DEC/2013Hajime Muramatsu U of Minnesota 1

Measurements ofstrong phase in D0 ➝ Kπ decay and yCPvia quantum-‐correla>ons at BESIII

Hajime Muramatsu, University of Minnesota(for the BESIII collabora>on)

-‐ Strong phase in D0 ➝ Kπ decay-‐ yCP measurement

Miami 2013 DEC/2013Hajime Muramatsu U of Minnesota

Beijing Electron Positron Collider (BEPC-‐II)-‐ A symmetric e+e-‐ collider, opera>ng at Ecm ~ 2.0 ~4.6 GeV (Charm factory!).

-‐ It’s in Beijing: Easy access to the downtown area of Beijing with a nearby subway sta>on!

2

LinacStorage ring

Where I sleep:Next to a Chinese restaurantCoun>ng room:

where I take shi\s

BESIII detector


BESIII detector• A powerful general purpose detector.

• Excellent neutral and charged par8cle detec8on and iden8fica8on with a large coverage.

3

The BESIII detector

Xiao-Rui Lu @ Charm 2013 6 6

The new BESIII detector is hermetic for neutral and charged particle with excellent resolution, PID, and large coverage.

NIM A614, 345 (2010)

Magnet: 1 T Super conducting MDC: small cell & He gas xy=130 m sp/p = 0.5% @1GeV dE/dx=6%

TOF: T = 90 ps Barrel 110 ps Endcap

Muon ID: 8~9 layer RPC RΦ=1.4 cm~1.7 cm

EMCAL: CsI crystal E/E = 2.5% @1 GeV φ,z = 0.5~0.7 cm/E

Trigger: Tracks & Showers Pipelined; Latency = 6.4 s

Data Acquisition: Event rate = 3 kHz Throughput ~ 50 MB/s


Data samples we have

4

-‐ @ J/ψ peak : 1.2 B J/ψ decays and some scan in the vicinity of the peak.

-‐ @ ψ(3686) peak : 0.5 B ψ(3686) decays and some scan in the vicinity of the peak.

-‐ Above DDP threshold: 0.5/Q @ Ecm = 4.009 GeV, 1.9/a @ Ecm = 4.26 GeV, 0.5/a @ Ecm = 4.36 GeV, plus some scan samples as well.

The above samples have been producing very rich Physics resultssuch as hadron spectroscopy of Charmonia (e.g., hc/ηc)and of Charmonium-‐like states (X/Y/Z).

-‐ Today, I report recent results from BESIII based on a sample that was taken near DDj threshold: 2.92/a @Ecm = 3.773 GeV


Sample at Ecm = 3.773 GeV

-‐ The total integrated luminosity of 2.92 Q-‐1 at this energy point is the largest in the world to date.

-‐ In the selected hadronic events (mul8ple reconstructed charged/ neutral hadrons or tracks), they are dominated by;

e+e-‐ ➝ γ* ➝ ψ(3770) and e+e-‐ ➝ γ* ➝ (qq̅) light hadrons in which σ(e+e-‐ ➝ ψ(3770) ➝ hadrons)/σ(e+e-‐ ➝ NR ➝ hadrons) ~ 1/2.

-‐ Once ψ(3770) is produced, it predominantly decays into a DDP pair.For instance, we have ~21 M D0 (or DP0) decays in this sample.

-‐ Rela8vely clean event environment.

-‐ When the two D mesons are reconstructed, the sample becomes almost background free.

5


Things can be done with the sample taken at or around Ecm = 3.773 GeV

-‐ There are many interes8ng possible topics to study in D (weak) decays based on our sample, such as;

-‐ pure leptonic decays(e.g., extrac8on of |Vcd| and/or its decay constant, fD).

-‐ Semi-‐leptonic decays(e.g., extrac8on of their form factors, and then compare them vs B meson case).

-‐ With the largest sample of D mesons taken at the near threshold, one should look for rare/forbidden decays (e.g., FCNC, LNV, LFV).

-‐ or even ψ(3770) itself such as ψ(3770) ➝ non-‐DDP final states.-‐ But today, I report our ahempt to measure some of the parameters of DDP mixing using the unique characteris8cs of our ψ(3770) data set taken at Ecm = 3.773 GeV.

6


Introduc>on-‐ DDP mixing is highly suppressed by the GIM mechanism and by the CKM matrix elements

within the Standard Model.

-‐ Observa8on of DDP mixing, first seen by theB factories (HFAG: arXiv 1207.1158) and now observedby LHCb: PRL110, 101802 (2013).

-‐ Improving the constraints on the charm mixingparameter is important for tes8ng the SM,such as long distance effects.

-‐ DDP mixing is conven8onally described by two parameters: x = 2(M1-‐M2)/(𝚪1+𝚪2), y = (𝚪1-‐𝚪2)/(𝚪1+𝚪2),where M1,2 and 𝚪1,2 are the masses and widths of the neutral D meson mass eigenstates.(Flavor eigenstates, D0/DP0, are not the same as mass eigenstates, D1/D2)Or x’ = x·∙cosδKπ + y·∙sinδKπ, y’ = y·∙cosδKπ -‐ x·∙sinδKπ.

-‐ δKπ is the strong phase difference between the doubly Cabibbo suppressed (DCS) decay, DP0 ➝ K-‐π+ and the Cabibbo favored (CF) decay, D0 ➝ K-‐π+ or ⟨K-‐π+∣DP0⟩/⟨K-‐π+∣D0⟩ = -‐r·∙e-‐iδ.So one can connect (x,y) with (x’,y’) via δKπ.

-‐ In this talk, I present preliminary results on δKπ and y using the quantum correla8onbetween the produced D0 and DP0 pair in data taken at BESIII.

7

The decay time, ti, is the average value in each bin of theRS sample. The fit parameters, !, include the three mixingparameters (RD, y

0, x02) and five nuisance parameters usedto describe the decay time evolution of the secondary Dfraction (!B) and of the peaking background (!p). Thenuisance parameters are constrained to the measured val-ues by the additional !2

B and !2p terms, which account for

their uncertainties including correlations.The analysis procedure is defined prior to fitting the data

for the mixing parameters. Measurements on pseudoex-periments that mimic the experimental conditions of thedata, and where D0 ! "D0 oscillations are simulated, indi-cate that the fit procedure is stable and free of any bias.

The fit to the decay-time evolution of the WS/RS ratio isshown in Fig. 2 (solid line), with the values and uncertain-ties of the parameters RD, y

0 and x02 listed in Table I. Thevalue of x02 is found to be negative but consistent with zero.As the dominant systematic uncertainties are treated withinthe fit procedure (all other systematic effects are negli-gible), the quoted errors account for systematic as well asstatistical uncertainties. When the systematic biases are notincluded in the fit, the estimated uncertainties on RD, y

0,and x02 become, respectively 6%, 10%, and 11% smaller,

showing that the quoted uncertainties are dominated bytheir statistical component. To evaluate the significance ofthis mixing result, we determine the change in the fit !2

when the data are described under the assumption of theno-mixing hypothesis (dashed line in Fig. 2). Under theassumption that the !2 difference, !!2, follows a !2

distribution for two degrees of freedom, !!2 ¼ 88:6 cor-responds to a p-value of 5:7# 10!20, which excludes theno-mixing hypothesis at 9.1 standard deviations. This isillustrated in Fig. 3 where the 1", 3", and 5" confidenceregions for x02 and y0 are shown.As additional cross-checks, we perform the measure-

ment in statistically independent subsamples of the data,selected according to different data-taking periods, andfind compatible results. We also use alternative decay-time binning schemes, selection criteria or fit methods toseparate signal and background, and find no significantvariations in the estimated parameters. Finally, to assessthe impact of events where more than one candidate isreconstructed, we repeat the time-dependent fit on dataafter randomly removing the additional candidates andselecting only one per event; the change in the measuredvalue of RD, y0, and x02 is 2%, 6%, and 7% of theiruncertainty, respectively.In conclusion, we measure the decay time dependence of

the ratio between D0 ! Kþ#! and D0 ! K!#þ decaysusing 1:0 fb!1 of data and exclude the no-mixing hypothe-sis at 9.1 standard deviations. This is the first observation ofD0 ! "D0 oscillations in a single measurement. The mea-sured values of the mixing parameters are compatible withand have substantially better precision than those fromprevious measurements [4,6,23].We express our gratitude to our colleagues in the CERN

accelerator departments for the excellent performance ofthe LHC. We thank the technical and administrative staff atthe LHCb institutes. We acknowledge support from CERNand from the national agencies: CAPES, CNPq, FAPERJ,

TABLE I. Results of the time-dependent fit to the data. Theuncertainties include statistical and systematic sources; ndfindicates the number of degrees of freedom.

Fit typeParameter

Fit result Correlation coefficient(!2=ndf) (10!3) RD y0 x02

Mixing RD 3:52% 0:15 1 !0:954 þ0:882(9:5=10) y0 7:2% 2:4 1 !0:973

x02 !0:09% 0:13 1No mixing RD 4:25% 0:04(98:1=12)

[%]2x'-0.1 -0.05 0 0.05

[%]

y'

-0.5

0

0.5

1

1.5

2

σ5

σ3

σ1

No-mixing

LHCb

FIG. 3. Estimated confidence-level (C.L.) regions in the(x02, y0) plane for 1! C:L: ¼ 0:317 (1"), 2:7# 10!3 (3"),and 5:73# 10!7 (5"). Systematic uncertainties are included.The cross indicates the no-mixing point.

τ/t0 2 4 6 20

R

3

3.5

4

4.5

5

5.5

6

6.5

7

-310×

DataMixing fit

No-mixing fit

LHCb

FIG. 2 (color online). Decay-time evolution of the ratio, R, ofWS D0 ! Kþ#! to RS D0 ! K!#þ yields (points) with theprojection of the mixing allowed (solid line) and no-mixing(dashed line) fits overlaid.

PRL 110, 101802 (2013) P HY S I CA L R EV I EW LE T T E R Sweek ending

8 MARCH 2013

101802-4


The decay rate of a correlated state‣ For physical process producing D0DP0 such as

e+e-‐ ➝ γ* ➝ ψ(3770) ➝ D0DP0, the D0DP0 pair are in a quantum-‐correlated state. The quantum number of ψ(3770) is JPC = 1-‐-‐. Thus, the D0DP0 pair in this process has C = -‐. For a correlated state with C = -‐, the two D mesons are an8-‐symmetric in the limit of CP invariance:

‣ The two produced neutral mesons must have opposite CP(i.e., see Goldhaber and Rosner, PRD15, 1254 (1977). That is;

‣ Final states of (CP+, CP+) or (CP-‐, CP-‐) are forbidden.

‣ Final states of (CP+, CP-‐) are maximally enhanced (doubled).

‣ Final states of CP± against inclusive states (Single tag or ST) are not affected.

‣ Final states of (K-‐π+, CP±) are affected due to the interference between CF and DCS (δKπ).

8

Production at threshold


Threshold production at 3.773 GeV Double Tag techniques: (partial-)reconstruct

both D mesons Charm events at threshold are very clean and

unique in studying D decays

Quantum correlation of two D mesons Very clean environment with little to

no non-DDbar background Lots of systematic uncertainties

uncertainties cancel when applying double tag method

Xiao-Rui Lu @ Charm 2013 4

The decay rate of a correlated state

Taking advantage the quantum coherence of DDbar pairs, BESIII can study the charm physics in an unique way • strong phase • mixing parameters • direct CP violation • ...


Extrac>ng δKπ‣ Neglec8ng higher orders in the mixing parameters (e.g., y2), one can arrive at the following rela8on: 2·∙r·∙cosδKπ+y = (1+RWS)·∙ACP➝Kπ,where RWS ≡ Γ(DP0➝K-‐π+)/Γ(D0➝K-‐π+) and ACP➝Kπ ≡ [B(D2➝K-‐π+) -‐ B(D1➝K-‐π+)]/B(D2➝K-‐π+) + B(D1➝K-‐π+)].

‣ We can extract ACP➝Kπ by tagging one D (tag side) with exclusive CP-‐eigenstates which then defines the eigenvalueof the other D (ACP± ≡ ⟨K-‐π+∣D1,2⟩).

‣ Then, with the knowledge of r, y, and RWS from the 3rd par8es (HFAG2013 and PDG), we could derive cosδKπ in the end.

‣ The rest of the analysis becomes measurements of B(DCP±➝K-‐π+) while simultaneously reconstruc8ng the DCP∓ on the tag side.

9

Accessing strong phase δKπ at threshold


We measure the strong phase difference using quantum correlated production of D-Dbar at the production threshold

based on 2.9 fb-1 ψ(3770) data

When we neglect CPV, CP of the two D mesons are anti-symmetric.


Measuring B(DCP±➝K-‐π+) • Double-‐Tag technique:B(DCP±➝Kπ) = [B(DCP∓➝ CP∓ states)×B(DCP±➝Kπ)]/B(DCP∓➝ CP∓ states) = (nKπ,CP∓/nCP∓)·∙(εCP∓/εKπ,CP∓),where nKπ,CP∓ are yields of “Kπ” when CP states are simultaneously reconstructed on the tag side nCP∓ are yields of CP states (independent of how the other D decays) εCP∓ and εKπ,CP∓ are the corresponding reconstruc8on efficiencies.

• “Yields” are extracted from Mbc distribu8ons :

• CP states on Tag side (8 modes):

where we reconstruct KS➝π+π-‐, π0/η➝γγ, ω➝π+π-‐π0, ρ➝π+π-‐.

• No8ce that most of systema8cs on the tag side get canceled in B(DCP±➝Kπ).The remaining systema8cs (reconstruc8on/simula8on) of Kπ are also canceled in the determina8on of ACP➝Kπ .

10

Single tags of CP modes


preliminary preliminary

preliminary preliminary preliminary


22DbeamBC pEM

unique variable in threshold-production

Type ModeFlavored K−π+, K+π−

CP+ K+K−, π+π−, K0Sπ

0π0, π0π0, ρ0π0

CP− K0Sπ

0, K0Sη, K

0Sω

Table 1: D decay modes reconstructed in the analysis of δKπ.

Here, nCP± (nKπ,CP±) and εCP± (εKπ,CP±) are yields and detection efficiencies of sin-gle tags (ST) of D → CP± (double tags (DT) of D → CP±, D → Kπ), respectively.With external inputs of the parameters of r, y and RWS, we can extract δKπ fromACP→Kπ. Based on a dataset of 818 pb−1 of collision data collected with the CLEO-cdetector at the center of mass

√s = 3.77GeV, the CLEO collaboration measured

cos δKπ = 0.81+0.22+0.07−0.18−0.05 [10]. Using a global fit method with inclusion of the external

mixing parameters, CLEO obtained cos δKπ = 1.15+0.19+0.00−0.17−0.08 [10].

We choose 5 CP -even D0 decay modes and 3 CP -odd modes, as listed in Tab. 1,with π0 → γγ, η → γγ, K0

S → π+π− and ω → π+π−π0. Variable

MBC ≡√E2

0/c4 − |(pD|2/c2

is plotted in Fig. 1 to identify the CP ST signals, where (pD is the total momentumof the D0 candidate and E0 is the beam energy. Yields of the CP ST signals areestimated by maximum likelihood fits to data, in which signal shapes are derivedfrom MC simulation convoluted with a smearing Gaussian function, and backgroundfunctions are modeled with the ARGUS function [11]. In the events of the CP STmodes, we reconstruct the Kπ combinations using the remaining charged tracks withrespect to the ST D candidates. Similar fits are implemented to the distributions ofMBC(D → CP±) in the survived DT events to estimate yields of DT signals. Thefits are shown in Fig. 2.

We get the asymmetry to be

ACP→Kπ = (12.77± 1.31+0.33−0.31)%,

where the first uncertainty is statistical and the second is systematic. To measure thestrong phase δKπ in Eq. (1), we quote the external inputs of RD = r2 = 3.47±0.06‰,y = 6.6±0.9‰, and RWS = 3.80±0.05‰ from HFAG 2013 [12] and PDG [5]. Hence,we obtain

cos δKπ = 1.03± 0.12± 0.04± 0.01,

where the first uncertainty is statistical, the second uncertainty is systematic, and thethird uncertainty is due to the errors introduced by the external input parameters.This result is more precise than CLEO’s measurement and provides the world bestconstrain to δKπ.

3


Yields of CP states (nCP∓)(reconstruct only one of the two neutral D)

11

)2(GeV/cBCM1.84 1.86 1.88

)2Ev

ents

/(1 M

eV/c

0

5000

10000

15000

)2(GeV/cBCM1.84 1.86 1.88

)2Ev

ents

/(1 M

eV/c

0

5000

10000

15000- K+ K

)2(GeV/cBCM1.84 1.86 1.88

)2Ev

ents

/(1 M

eV/c

0

500

1000

1500

)2(GeV/cBCM1.84 1.86 1.88

)2Ev

ents

/(1 M

eV/c

0

500

1000

1500 0π0π

)2(GeV/cBCM1.84 1.86 1.88

)2Ev

ents

/(1 M

eV/c

0

5000

10000

)2(GeV/cBCM1.84 1.86 1.88

)2Ev

ents

/(1 M

eV/c

0

5000

100000πS

0 K

)2(GeV/cBCM1.84 1.86 1.88

)2Ev

ents

/(1 M

eV/c

0

2000

4000

6000

)2(GeV/cBCM1.84 1.86 1.88

)2Ev

ents

/(1 M

eV/c

0

2000

4000

6000-π+π

)2(GeV/cBCM1.84 1.86 1.88

)2Ev

ents

/(1 M

eV/c

0

5000

10000

)2(GeV/cBCM1.84 1.86 1.88

)2Ev

ents

/(1 M

eV/c

0

5000

100000πρ

)2(GeV/cBCM1.84 1.86 1.88

)2Ev

ents

/(1 M

eV/c

0

500

1000

1500

2000

)2(GeV/cBCM1.84 1.86 1.88

)2Ev

ents

/(1 M

eV/c

0

500

1000

1500

2000ηS

0 K

)2(GeV/cBCM1.84 1.86 1.88

)2Ev

ents

/(1 M

eV/c

0

2000

4000

)2(GeV/cBCM1.84 1.86 1.88

)2Ev

ents

/(1 M

eV/c

0

2000

40000π0πS

0 K

)2(GeV/cBCM1.84 1.86 1.88

)2Ev

ents

/(1 M

eV/c

0

2000

4000

6000

8000

)2(GeV/cBCM1.84 1.86 1.88

)2Ev

ents

/(1 M

eV/c

0

2000

4000

6000

8000ωS

0 K

Figure 1: ST MBC distributions of the D → CP± decays and fits to data. Data areshown in points with error bars. The solid lines show the total fits and the dashedlines show the background shapes.

4

- Signal shape: MC shape, convoluted with a Gaussian (to compensate the difference in resolu>on between data and MC).

- Background: ARGUS background func>on.Prelimin

aryCP+

CP+

CP-‐


Can also check “CP purity”- When D0 and Dj0 are reconstructed, final states with the same CP should yield zero

events.

12


CP purity check of CP-tag modes



preliminary

events with same-CP decays are consistent with 0

consider as systematic uncertainty


CP purity check of CP-tag modes



preliminary

events with same-CP decays are consistent with 0

consider as systematic uncertainty

✴ Consistent with zero.

✴ Consider as one of the systema>cs.

CP+ CP-‐

CP+CP+CP-‐


Yields of Kπ in double tags (nKπ,CP∓)(reconstruct CP-‐final state from one D decay,

with “Kπ” from the other D)

13

)2(GeV/cBCM1.84 1.86 1.88

)2Ev

ents

/(2 M

eV/c

0

500

1000

)2(GeV/cBCM1.84 1.86 1.88

)2Ev

ents

/(2 M

eV/c

0

500

1000- K+, Kπ K

)2(GeV/cBCM1.84 1.86 1.88

)2Ev

ents

/(2 M

eV/c

0

20

40

60

80

)2(GeV/cBCM1.84 1.86 1.88

)2Ev

ents

/(2 M

eV/c

0

20

40

60

800π0π,π K

)2(GeV/cBCM1.84 1.86 1.88

)2Ev

ents

/(2 M

eV/c

0

200

400

600

)2(GeV/cBCM1.84 1.86 1.88

)2Ev

ents

/(2 M

eV/c

0

200

400

600 0πS0, Kπ K

)2(GeV/cBCM1.84 1.86 1.88

)2Ev

ents

/(2 M

eV/c

0

100

200

300

)2(GeV/cBCM1.84 1.86 1.88

)2Ev

ents

/(2 M

eV/c

0

100

200

300-π+π,π K

)2(GeV/cBCM1.84 1.86 1.88

)2Ev

ents

/(2 M

eV/c

0

200

400

)2(GeV/cBCM1.84 1.86 1.88

)2Ev

ents

/(2 M

eV/c

0

200

4000πρ,π K

)2(GeV/cBCM1.84 1.86 1.88

)2Ev

ents

/(2 M

eV/c

0

50

100

)2(GeV/cBCM1.84 1.86 1.88

)2Ev

ents

/(2 M

eV/c

0

50

100η

S0, Kπ K

)2(GeV/cBCM1.84 1.86 1.88

)2Ev

ents

/(2 M

eV/c

0

100

200

)2(GeV/cBCM1.84 1.86 1.88

)2Ev

ents

/(2 M

eV/c

0

100

200

0π0πS0, Kπ K

)2(GeV/cBCM1.84 1.86 1.88

)2Ev

ents

/(2 M

eV/c

0

100

200

300

)2(GeV/cBCM1.84 1.86 1.88

)2Ev

ents

/(2 M

eV/c

0

100

200

300 ωS0, Kπ K

Figure 2: DT MBC distributions and the corresponding fits. Data are shown inpoints with error bars. The solid lines show the total fits and the dashed lines showthe background shapes.

5

- Signal shape: MC shape, convoluted with a Gaussian (to compensate for the difference in resolu>on between data and MC).

- Background: ARGUS background func>on.Prelimin

ary


Preliminary fit results

- These yields allow us to obtain B(DCP±➝K-‐π+) which then provides ACP➝Kπ.- ACP➝Kπ = (12.77±1.31(stat.)+0.33-‐0.31(syst.))%.

14


Preliminary numerical results

Prelimin

ary


Preliminary result on δKπ- We have measured ACP➝Kπ = (12.77±1.31(stat.)+0.33-‐0.31(syst.))%.- Using the rela>on, 2·∙r·∙cosδKπ+y = (1+RWS)·∙ ACP➝Kπ,and with external inputs from HFAG2013 and PDG(RD = 3.47±0.06%, y=6.6±0.9%, RWS = 3.80±0.05%),we obtain

cosδKπ = 1.03±0.12(stat.)±0.04(syst.)±0.01(external).

- Our result is consistent with and more precise than the recent CLEO result (PRD86, 112001 (2012)): cosδKπ = 1.15+0.19-‐0.17(stat.)+0.00-‐0.08(syst.).

15


Determina>on of the mixing parameter, yCP

- yCP is defined as; 2·∙yCP = (|q/p|+|p/q|)·∙y·∙cosφ -‐(|q/p|-‐|p/q|)·∙x·∙sinφ,where p and q are mixing parameters,and φ = arg(q/p) is the weak phase difference of themixing amplitudes.No>ce: for no CPV case, p = q = 1/√2 and yCP ≡ y.

- For D decays into any CP-‐eigenstate, its decay rate can be described as; R(D0/Dj0 ➝ CP±) ∝ |ACP±|2 ·∙ (1∓yCP).

- When one D decays into a CP-‐eigenstate, while the other D decays semi-‐leptonically, the decay rate can be given by; R(D0/Dj0 ➝ CP±, and Dj0/D0 ➝ semi-‐lep ) ∝ |Al|2 ·∙|ACP±|2.- Semileptonic decay width does not depend on the CP of its parent D.

- Yet, the total width of its parent D depends on CP.

- Result: semileptonic BF of D1,2 gets modified by a factor of 1±yCP.

- Combining the above two, and neglec>ng terms with y2 (or higher), one can arrive at

16

Type ModesCP+ K+K−, π+π−, KSπ0π0

CP− K0Sπ

0, K0Sω, K

0Sη

l± Keν, Kµν

Table 2: CP -tag modes and D semileptonic decay modes.

3 Measurement of yCP

For D decays to any CP -eigenstate final states, their decay rates can be formulatedto be

R(D0/D0 → CP±) ∝ |ACP± |2(1∓ yCP ).

When the partner D decays semileptonically in the DD pair production at threshold,double decay rates will be

Rl;CP± = |Al|2|ACP± |2.If we take the ratio of the single decay rate and the double decay rate and neglecthigher order of y2CP, we can extract yCP using the following equation [4]

yCP =1

4(Rl;CP+RCP−

Rl;CP−RCP+− Rl;CP−RCP+

Rl;CP+RCP−).

Similar to the notations in Eq. (3), experimentally we denote the decay rate ratios ofRl;CP±RCP±

to be B± and determine it with the D tagging method

B± =nl;CP±

nCP±· εCP±

εl;CP±.

Hence, yCP = 14 [

B̃+

B̃−− B̃−

B̃+], where B̃± is combinations of different CP -tag mode α

using the least square method

χ2 =∑

α

(B̃± −Bα±)

2

(σα±)2

.CP -tag modes in Tab. 2 are used in this analysis. Similar to the analysis of

δKπ, ST yields are estimated by fits to the MBC distributions, as shown in Fig. 3.Semileptonic decays of D → Keν and D → Kµν are selected with respect to theCP -tagged D candidates in ST events. Due to the undetectable neutrino in the finalstates, variable Umiss is used to distinguish the signals of semileptonic decays frombackgrounds. The definition is given as

Umiss ≡ Emiss − |*pmiss|,

6

- The expression for yCP can be wri�en as;

where B*± is the branching frac>on, averaged over different CP tag modes, α, that is obtained by minimizing

- All branching frac>ons are obtained in a similar way, the double-‐tag method.

- When the semileptonic decays are reconstructed, however, we use Umiss distribu>ons to obtain their yields, instead of Mbc, ,which peaks ~ 0 if only missing par>cle is neutrino.

- Tag modes:


Extrac>ng yCP in our experiment17


CP− K0Sπ

0, K0Sω, K

0Sη

l± Keν, Kµν




R(D0/D0 → CP±) ∝ |ACP± |2(1∓ yCP ).



yCP =1

4(Rl;CP+RCP−


Rl;CP+RCP−).



B± =nl;CP±

nCP±· εCP±

εl;CP±.

Hence, yCP = 14 [

B̃+

B̃−− B̃−



χ2 =∑

α

(B̃± −Bα±)

2

(σα±)2




6


CP− K0Sπ

0, K0Sω, K

0Sη

l± Keν, Kµν




R(D0/D0 → CP±) ∝ |ACP± |2(1∓ yCP ).



yCP =1

4(Rl;CP+RCP−


Rl;CP+RCP−).



B± =nl;CP±

nCP±· εCP±

εl;CP±.

Hence, yCP = 14 [

B̃+

B̃−− B̃−



χ2 =∑

α

(B̃± −Bα±)

2

(σα±)2




6


CP− K0Sπ

0, K0Sω, K

0Sη

l± Keν, Kµν




R(D0/D0 → CP±) ∝ |ACP± |2(1∓ yCP ).



yCP =1

4(Rl;CP+RCP−


Rl;CP+RCP−).



B± =nl;CP±

nCP±· εCP±

εl;CP±.

Hence, yCP = 14 [

B̃+

B̃−− B̃−



χ2 =∑

α

(B̃± −Bα±)

2

(σα±)2




6Type ModesCP+ K+K−, π+π−, KSπ0π0

CP− K0Sπ

0, K0Sω, K

0Sη

l± Keν, Kµν




R(D0/D0 → CP±) ∝ |ACP± |2(1∓ yCP ).



yCP =1

4(Rl;CP+RCP−


Rl;CP+RCP−).



B± =nl;CP±

nCP±· εCP±

εl;CP±.

Hence, yCP = 14 [

B̃+

B̃−− B̃−



χ2 =∑

α

(B̃± −Bα±)

2

(σα±)2




6


Yields of CP states (nCP∓)(reconstruct only one of the two neutral D)

18

)2(GeV/cBCM1.84 1.86 1.88

2Ev

ents

/0.6

MeV

/c

0

5000

10000

)2(GeV/cBCM1.84 1.86 1.88

2Ev

ents

/0.6

MeV

/c

0

5000

10000- K+K

)2(GeV/cBCM1.84 1.86 1.88

2Ev

ents

/1.0

MeV

/c0

1000

2000

3000

4000

)2(GeV/cBCM1.84 1.86 1.88

2Ev

ents

/1.0

MeV

/c0

1000

2000

3000

4000-π +π

)2(GeV/cBCM1.84 1.86 1.88

2Ev

ents

/ 1.

0 M

eV/c

0

2000

4000

)2(GeV/cBCM1.84 1.86 1.88

2Ev

ents

/ 1.

0 M

eV/c

0

2000

4000

)2(GeV/cBCM1.84 1.86 1.88

2Ev

ents

/1.0

MeV

/c

0

5000

10000

)2(GeV/cBCM1.84 1.86 1.88

2Ev

ents

/1.0

MeV

/c

0

5000

100000π 0

SK

)2(GeV/cBCM1.84 1.86 1.88

2Ev

ents

/ 1.

0 M

eV/c

0

2000

4000

)2(GeV/cBCM1.84 1.86 1.88

2Ev

ents

/ 1.

0 M

eV/c

0

2000

4000 ω 0S K

)2(GeV/cBCM1.84 1.86 1.88

2Ev

ents

/1.0

MeV

/c

0

500

1000

1500

2000

)2(GeV/cBCM1.84 1.86 1.88

2Ev

ents

/1.0

MeV

/c

0

500

1000

1500

2000η 0

SK

Figure 3: MBC distributions and fits to data.

Emiss ≡ E0 − EK − El, !pmiss ≡ −[!pK + !pl + p̂ST

√E2

0 −m2D].

Here, EK/l (!pK/l) is the energy (three-momentum) of K± or lepton l∓, p̂ST is the unitvector in the reconstructed direction of the CP -tagged D and mD is the nominal D0

mass. The Umiss distributions are plotted in Fig. 4 for D → Keν and D → Kµνmodes.

In fits of the DT Keν modes, signal shape is modeled using MC shape convolutedwith an asymmetric Gaussian and backgrounds are described with a 1st-order polyno-mial function. In fits of the DT Kµν modes, signal shape is modeled using MC shapeconvoluted with an asymmetric Gaussian. Backgrounds of Keν are modeled usingMC shape and their relative rate to the signals are fixed. Shape of Kππ0 background-s are taken from MC simulations with convolution of a smearing Gaussian function;parameters of the smearing function are fixed according to fits to the control sampleof D → Kππ0 events. Size of Kππ0 backgrounds are fixed by scaling the number ofKππ0 events in the control sample to the number in the signal region according tothe ratio estimated from MC simulations. Other backgrounds are described with a1st-order polynomial function.

Finally, we obtain the preliminary result as

yCP = −1.6%± 1.3%(stat.)± 0.6%(syst.).

The result is compatible with the previous measurements [12]. This is the most precisemeasurement of yCP based on D0D0 threshold productions. However, its precision isstill statistically limited.

7

K+K-‐ π+π-‐ KSπ+π-‐

KSπ0 KSω KSη

Prelimin

aryCP+

CP-‐


Yields of Keν in double tags (nKeν,CP∓)(reconstruct CP-‐final states from one D decay,

with “Keν” from the other D)

19

(GeV)missU-0.1 -0.05 0 0.05 0.1

Even

ts /

( 0.0

05 )

0

50

100

150

200

(GeV)missU-0.1 -0.05 0 0.05 0.1

Even

ts /

( 0.0

05 )

0

50

100

150

200ν, Ke-K+K

(GeV)missU-0.1 -0.05 0 0.05 0.1

Even

ts /

( 0.0

05 )

0

20

40

60

80

(GeV)missU-0.1 -0.05 0 0.05 0.1

Even

ts /

( 0.0

05 )

0

20

40

60

80

ν, Ke-π+π

(GeV)missU-0.1 -0.05 0 0.05 0.1

Even

ts /

( 0.0

05 )

0

20

40

60

80

(GeV)missU-0.1 -0.05 0 0.05 0.1

Even

ts /

( 0.0

05 )

0

20

40

60

80

ν, Ke0π0π0SK

(GeV)missU-0.1 -0.05 0 0.05 0.1

Even

ts /

( 0.0

05 )

0

50

100

150

200

(GeV)missU-0.1 -0.05 0 0.05 0.1

Even

ts /

( 0.0

05 )

0

50

100

150

200ν, Ke0π0

SK

(GeV)missU-0.1 -0.05 0 0.05 0.1

Even

ts /

( 0.0

05 )

0

20

40

60

80

(GeV)missU-0.1 -0.05 0 0.05 0.1

Even

ts /

( 0.0

05 )

0

20

40

60

80

ν, Keω0SK

(GeV)missU-0.1 -0.05 0 0.05 0.1

Even

ts /

( 0.0

1 )

0

20

40

60

(GeV)missU-0.1 -0.05 0 0.05 0.1

Even

ts /

( 0.0

1 )

0

20

40

60ν, Keη0

SK

(GeV)missU-0.1 -0.05 0 0.05 0.1

Even

ts /

( 0.0

05 )

0

50

100

150

200

(GeV)missU-0.1 -0.05 0 0.05 0.1

Even

ts /

( 0.0

05 )

0

50

100

150

200νµ, K-K+K

(GeV)missU-0.1 -0.05 0 0.05 0.1

Even

ts /

( 0.0

05 )

0

20

40

60

80

(GeV)missU-0.1 -0.05 0 0.05 0.1

Even

ts /

( 0.0

05 )

0

20

40

60

80νµ, K-π+π

(GeV)missU-0.1 -0.05 0 0.05 0.1

Even

ts /

( 0.0

05 )

0

50

100

(GeV)missU-0.1 -0.05 0 0.05 0.1

Even

ts /

( 0.0

05 )

0

50

100

νµ, K0π0π0SK

(GeV)missU-0.1 -0.05 0 0.05 0.1

Even

ts /

( 0.0

05 )

0

100

200

(GeV)missU-0.1 -0.05 0 0.05 0.1

Even

ts /

( 0.0

05 )

0

100

200 νµ, K0π0SK

(GeV)missU-0.1 -0.05 0 0.05 0.1

Even

ts /

( 0.0

1 )

0

50

100

150

(GeV)missU-0.1 -0.05 0 0.05 0.1

Even

ts /

( 0.0

1 )

0

50

100

150νµ, Kω0

SK

(GeV)missU-0.1 -0.05 0 0.05 0.1

Even

ts /

( 0.0

1 )

0

20

40

60

(GeV)missU-0.1 -0.05 0 0.05 0.1

Even

ts /

( 0.0

1 )

0

20

40

60 νµ, Kη0SK

Figure 4: Fits to Umiss distributions in data for CP -tagged Keν and Kµν modes.

4 Summary

In this paper, the preliminary BESIII results of the strong phase difference cos δKπ

in D → Kπ decays and the mixing parameter yCP are reported. The measurementswere carried out based on the quantum-correlated technique in studying the processof D0D0 pair productions of 2.92 fb−1 e+e− collision data collected with the BESIIIdetector at

√s = 3.773GeV. The preliminary results are given as

cos δKπ = 1.03± 0.12± 0.04± 0.01

andyCP = −1.6%± 1.3%± 0.6%.

8

- Signal shape: MC shape, convoluted with an asymmetric Gaussian.

- Background: A 1st order polynomial.

K+K-‐, Keν π+π-‐, Keν KSπ+π-‐, Keν

KSπ0, Keν KSω, Keν KSη, Keν

Prelimin

ary


Yields of Kμν in double tags (nKμν,CP∓)(reconstruct CP-‐final states from one D decay,

with “Kμν” from the other D)

20

- Signal shape: MC shape, convoluted with an asymmetric Gaussian.

- Background: A 1st order polynomial. Kππ0 (dominant).

(GeV)missU-0.1 -0.05 0 0.05 0.1

Even

ts /

( 0.0

05 )

0

50

100

150

200

(GeV)missU-0.1 -0.05 0 0.05 0.1

Even

ts /

( 0.0

05 )

0

50

100

150

200ν, Ke-K+K

(GeV)missU-0.1 -0.05 0 0.05 0.1

Even

ts /

( 0.0

05 )

0

20

40

60

80

(GeV)missU-0.1 -0.05 0 0.05 0.1

Even

ts /

( 0.0

05 )

0

20

40

60

80

ν, Ke-π+π

(GeV)missU-0.1 -0.05 0 0.05 0.1

Even

ts /

( 0.0

05 )

0

20

40

60

80

(GeV)missU-0.1 -0.05 0 0.05 0.1

Even

ts /

( 0.0

05 )

0

20

40

60

80

ν, Ke0π0π0SK

(GeV)missU-0.1 -0.05 0 0.05 0.1

Even

ts /

( 0.0

05 )

0

50

100

150

200

(GeV)missU-0.1 -0.05 0 0.05 0.1

Even

ts /

( 0.0

05 )

0

50

100

150

200ν, Ke0π0

SK

(GeV)missU-0.1 -0.05 0 0.05 0.1

Even

ts /

( 0.0

05 )

0

20

40

60

80

(GeV)missU-0.1 -0.05 0 0.05 0.1

Even

ts /

( 0.0

05 )

0

20

40

60

80

ν, Keω0SK

(GeV)missU-0.1 -0.05 0 0.05 0.1

Even

ts /

( 0.0

1 )

0

20

40

60

(GeV)missU-0.1 -0.05 0 0.05 0.1

Even

ts /

( 0.0

1 )

0

20

40

60ν, Keη0

SK

(GeV)missU-0.1 -0.05 0 0.05 0.1

Even

ts /

( 0.0

05 )

0

50

100

150

200

(GeV)missU-0.1 -0.05 0 0.05 0.1

Even

ts /

( 0.0

05 )

0

50

100

150

200νµ, K-K+K

(GeV)missU-0.1 -0.05 0 0.05 0.1

Even

ts /

( 0.0

05 )

0

20

40

60

80

(GeV)missU-0.1 -0.05 0 0.05 0.1

Even

ts /

( 0.0

05 )

0

20

40

60

80νµ, K-π+π

(GeV)missU-0.1 -0.05 0 0.05 0.1

Even

ts /

( 0.0

05 )

0

50

100

(GeV)missU-0.1 -0.05 0 0.05 0.1

Even

ts /

( 0.0

05 )

0

50

100

νµ, K0π0π0SK

(GeV)missU-0.1 -0.05 0 0.05 0.1

Even

ts /

( 0.0

05 )

0

100

200

(GeV)missU-0.1 -0.05 0 0.05 0.1

Even

ts /

( 0.0

05 )

0

100

200 νµ, K0π0SK

(GeV)missU-0.1 -0.05 0 0.05 0.1

Even

ts /

( 0.0

1 )

0

50

100

150

(GeV)missU-0.1 -0.05 0 0.05 0.1

Even

ts /

( 0.0

1 )

0

50

100

150νµ, Kω0

SK

(GeV)missU-0.1 -0.05 0 0.05 0.1

Even

ts /

( 0.0

1 )

0

20

40

60

(GeV)missU-0.1 -0.05 0 0.05 0.1

Even

ts /

( 0.0

1 )

0

20

40

60 νµ, Kη0SK

Figure 4: Fits to Umiss distributions in data for CP -tagged Keν and Kµν modes.

4 Summary

In this paper, the preliminary BESIII results of the strong phase difference cos δKπ

in D → Kπ decays and the mixing parameter yCP are reported. The measurementswere carried out based on the quantum-correlated technique in studying the processof D0D0 pair productions of 2.92 fb−1 e+e− collision data collected with the BESIIIdetector at

√s = 3.773GeV. The preliminary results are given as

cos δKπ = 1.03± 0.12± 0.04± 0.01

andyCP = −1.6%± 1.3%± 0.6%.

8

K+K-‐, Kμν π+π-‐, Kμν KSπ+π-‐, Kμν

KSπ0, Kμν KSω, Kμν KSη, Kμν

Prelimin

aryKππ0 -‐ Kππ0 shapes and sizes

are fixed based oncontrol samples of actual data.-‐ The control samples are obtained by the same CP states and Kππ0, while ignoring the two photons from π0 decays to calculate Umiss.See the next slide for detail.


Fixing the Kππ0 shape- Obtain Eextra ≡ Sum of the all un-‐used energiesdeposited in EM calorimeter.- Eextra tends to be larger if it is Kππ0 due to the ignored extra photons from π 0 decay and is small if it is Kμν.- We actually do require Eextra<0.2 GeV to select Kμν signal candidates.

21

In the fit to Umiss, it is necessary to have a good understanding of Kππ0 back-grounds(Its resolution and fraction). In order to demonstrate this understanding,one can select pure Kππ0 to see Umiss distribution. In addition, due to the resolutionof Umiss is correlated to the tag side, it’s better to use the same tag modes. Thus,our effort was focus on selecting pure Kππ0 against CP tagged events.

We choose cut on the Eextra which is shown in Fig 13. Here we also show the{Umiss : Eextra} two dimensional plot in Fig 15. From which we see that the Umiss

is basically independent with the Eextra. We choose “Eextra > 0.5 GeV ” to selectthe pure Kππ0, then use the Umiss shape of Kππ0 in “Eextra > 0.5 GeV ” region todemonstrate Umiss shape in “Eextra < 0.2 GeV ” region.

(GeV)missU-0.1 -0.05 0 0.05 0.1

(GeV)

extra

E

0

0.2

0.4

0.6

0.8

1all

νµKK,K0ππKK,KνKK,Ke

others

(a) {Umiss : Eextra} two dimensional plot

(GeV)missU-0.1 -0.05 0 0.05 0.1

0

0.02

0.04

0.06

<0.2extraE>0.5extraE

0ππ, K-K+K

(b) Umiss distribution of Kππ0 events inEextra >0.5 GeV region and in Eextra <0.2GeV region

Figure 15: {Umiss : Eextra} two dimensional plot(a) (from MC) and Umiss distributionof Kππ0(b) from MC.

When “Eextra > 0.5 GeV ” was required, the resultant Umiss distributions forKππ0 from inclusive MC are shown in Fig. 16. Kππ0 is not very pure but with someother backgrounds. But the resultant Umiss distributions from data could be fittedusing MC shape, then we could get the resolution difference in data and MC. Weuse PDF from MC to describe flat backgrounds. Fit plots are shown in Fig 17. Theparameters of smearing Gaussian(which describe the resolution differences betweendata and MC) were listed in Table 9.

Through the fit, we could also get the number of Kππ0 events in “Eextra >0.5 GeV ” region. We define the “ratio” as the number of Kππ0 events in “Eextra <0.2 GeV ” region to the number in “Eextra > 0.5 GeV ” region. Then the number ofKππ0 events for data could be estimated using : NE<0.2

Kππ0 = ratio × NE>0.5Kππ0 . Where

ratios could be obtained from MC. We listed these numbers in Table 10.Then, in the Umiss fit, Kππ0 could be constrained well. The number of Kππ0

events could be fixed and parameters of its smearing function could also be fixed.Umiss fit plots are shown in Fig 18.

23

-‐ Fit to Umiss in Eextra>0.5 GeV where Kμν peak is suppressed.-‐ The fihed shape ≡ MC shape, convoluted with a Gaussian.

(Kππ0 yields in data in Eextra<0.2 GeV) = R×(Kππ0 yields in data in Eextra>0.5 GeV),where R = (Kππ0 yields in MC in Eextra<0.2 GeV)/(Kππ0 yields in MC in Eextra>0.5 GeV).

Fix shape

Fix size


Preliminary results- Fi�ed yields for each mode:

- A\er correc>ng for efficiencies (branching frac>ons), we arrive at yCP = [-‐1.6±1.3(stat.)±0.6(syst.)]%.- The result is sta>s>cally limited.- The systema>c uncertainty mainly comes from fi�ng procedures.

22

Preliminary numerical results


Signal yields of the full data set

preliminary result:

• result is statistically limited • systematic uncertainty is relatively small


Comparison with other measurements

- Our result is consistent with the world average (HFAG2013; this preliminary result is not included in the average).

- Also consistent with the latest result from CLEO-‐c (PRD86, 112001 (2012)); yCP = (4.2±2.0±1.0)%.(not listed in the figure).

23

Comparison with world measurement


compatible with world average results

BESIII (pre.) -1.6 ±1.3 ±0.6 %

best precision in Charm factory

World average directly from HFAG2013 (BESIII (pre.) not included)

CLEOc 2012: [PRD 86 (2012) 112001]

yCP=(4.2±2.0 ±1.0)%


Summary- Quantum-‐correlated D0Dj0 in e+e-‐ annihila>ons near threshold: Unique way to measure the Charm mixing parameters.

- Most precise measurement of strong phase difference in D0➝ Kπ.Will improve the determina>on of mixing parameters, x and y.

- Measurement of yCP: Sta>s>cally limited, consistent with the world average.

- Will collect larger “open-‐charm” data samples in years to come:Expect many interes>ng results.

24

Measurements*of strongphaseinD Kπdecayandy...

Documents

Transcript of Measurements*of strongphaseinD Kπdecayandy...